CSO-LLM: Class Subspace Orthogonalization for Post-Training Backdoor Detection and Trigger Inversion in LLMs

Zhengxing Li, David J. Miller, Guangmingmei Yang +1

While post-training backdoor detection and trigger inversion schemes have been developed for AIs used e.g. for images, there is a paucity of such methods for LLMs. First, the LLM i…